How To Calculate Number Of Accidents From Accident Raw Data Of Accident Id
Int J Environ Res Public Health. 2020 Feb; 17(4): 1393.
Relationship Between Traffic Volume and Accident Frequency at Intersections
Received 2020 January 31; Accepted 2020 Feb xix.
Abstruse
Driven by the high social costs and emotional trauma that outcome from traffic accidents effectually the world, research into understanding the factors that influence accident occurrence is critical. There is a lack of consensus nigh how the management of congestion may affect traffic accidents. This paper aims to improve our agreement of this relationship past analysing accidents at 120 intersections in Adelaide, Australia. Data comprised of 1629 motor vehicle accidents with traffic volumes from a dataset of more than than 5 million hourly measurements. The effect of rainfall was also examined. Results showed an approximately linear relationship betwixt traffic volume and accident frequency at lower traffic volumes. In the highest traffic volumes, poisson and negative binomial models showed a significant quadratic explanatory term as accident frequency increases at a higher rate. This implies that focusing management efforts on avoiding these conditions would exist most effective in reducing blow frequency. The relative risk of rainfall on blow frequency decreases with increasing congestion index. Accident chance is five times greater during pelting at low congestion levels, successively decreasing to no elevated take a chance at the highest congestion level. No significant effect of congestion index on accident severity was detected.
Keywords: traffic volume, congestion, intersections, rainfall gamble, relative risk, urban
one. Introduction
Traffic accidents were estimated to cost Australia A$33.15 billion in 2016 [one]. This figure comes from estimations of the "value of a statistical life" [2] and costs associated with loss of economic output equally a issue of injury as well every bit the repair of belongings [3]. In the Usa, fatalities from traffic accidents surpassed the combined toll taken by the two most mortiferous diseases, cancer and heart illness, and close to half of the deaths of 19-twelvemonth-olds were a issue of traffic accidents [four]. While almanac route fatalities per 100,000 people in Commonwealth of australia were 5 times less in 2013 compared to 1975 [five], in that location were over 1100 fatalities in 2018 [6]. To maintain this reduction in fatalities and reduce accidents overall, understanding the range of causative factors that influence traffic accidents is critical.
Influencing factors include environmental conditions [7,8,9,10,11,12,xiii], vehicle factors [four,fourteen], driver characteristics and behaviour [15,16,17,xviii], and road blueprint [nineteen,20].
Of particular involvement is the effect of traffic weather on accidents. While on the surface it seems desirable to reduce congestion, if information technology correlates negatively with serious injury or fatal accident frequency, a reduction may negatively impact road safety [21]. A strong understanding of this relationship is necessary to improve traffic management and reduce accident frequency. Research stems from the 1930s [eleven], with relationships betwixt accident occurrence and traffic book/congestion falling into one of two wide categories: linear and non-linear [22].
Veh [11] establish a positive correlation between accident rates and average daily traffic (ADT), before accident rates gradually declined in higher traffic volumes; a tendency besides found past Raff [23]. Other studies using ADT and annual boilerplate daily traffic (AADT) reported uncomplicated positive linear correlations [24,25,26,27,28]. Gwynn [29] suggested that the higher temporal resolution of hourly traffic information may give a stronger relationship. Using hourly volumes, both Gwynn [29] and Ceder [30] institute a U-shaped curve, with the highest accident rates existing in the lowest and highest traffic volumes. Martin [31] likewise constitute a U-shaped response, as did Frantzeskakis and Iordanis [32] when because the relationship to level of service. Shefer [33] hypothesized that the human relationship between the volume/chapters (v/c) ratio and fatal accident frequency would form a bell-shaped curve. This hypothesis was supported past Martin [31] when looking at overall accident frequencies (not only fatal) and hourly traffic when using half-dozen-minute traffic book measurements for periods when accidents occurred.
Fortunately, accidents are rare events. Notwithstanding, from a statistical perspective, this requires analysis over long time periods and broad spatial scales to ensure sufficient sample sizes. Large datasets better the sensitivity of model response to variables of interest. Advanced modelling approaches use high temporal resolution traffic volume information in combination with multiple covariates to predict accident frequencies, every bit in a written report past Theofilatos [34]. Merely these approaches cannot be hands adopted elsewhere if the information required to include these covariates is non bachelor (e.g., road geometry, moisture conditions or light levels) and more parsimonious models may be necessary [35]. It is important to notation that a parsimonious approach could lead to bug relating to unobserved heterogeneity in unincluded factors [35] between different accidents and intersections. Approaches such as the v/c ratio exist to standardise traffic volume with relation to intersection capacity [36] and allow detailed assay at a limited number of locations past taking differences in intersection characteristics into business relationship. Just the application of this method across broad spatial scales is difficult if road geometry, directional traffic book, or traffic signal data is unavailable.
This study aims to analyse how traffic volumes affect accident frequency to address the lack of consensus between the linear and not-linear hypotheses in the wake of by research. Large datasets of high temporal frequency traffic volumes are used and the response of accident occurrence to congestion across 120 intersections will exist analysed. Split analyses look at the effect of congestion on accident severity and the effects of rainfall on accident risk beyond these congestion levels. The City of Adelaide is chosen due to overlap in availability of high temporal frequency, spatially explicit traffic data and detailed accident records.
two. Materials and Methods
This analysis combines detailed spatio-temporal traffic accident records and hourly intersection traffic volume information. Past normalising traffic volumes to each intersection, the resulting congestion alphabetize allowed traffic conditions to be compared between intersections irrespective of differences in intersection characteristics.
Ii farther factors were investigated. Accidents risks at different congestion levels were analysed with relation to rainfall and accidents were disaggregated by severity level to uncover whatsoever influence of congestion on blow severity.
Data was processed and analysed using the R programming linguistic communication [37] with the RStudio integrated evolution environment [38].
2.1. Study Area
The study is constrained to the Adelaide City Council (ACC) area in South Australia, Australia—called based on the extent of the hourly intersection traffic volumes dataset. Figure i shows the location relative to wider Adelaide and Australia as a whole.
Location of the study area and intersection sites where high temporal resolution traffic book data exists.
2.2. Information Processing Workflow
Effigy 2 summarises the process past which intersection traffic volumes were joined to traffic blow records with reference to the relevant methods sections.
Workflow for processing and joining the intersection traffic volumes and traffic accident datasets.
2.three. Accident Data
Traffic blow records were obtained from the Department of Planning, Transport and Infrastructure'southward (DPTI) "Road Crash Data" dataset [39]. While this is publicly available, the dates and times of individual accidents are omitted for privacy reasons. The DPTI provided a version with dates and times included for use in this research. The dataset contains information about each accident, including the date and fourth dimension, coordinates, weather conditions and accident severity.
A separate "units" tabular array provides boosted information most the units (including cars, cyclists and fixed objects) involved in each accident.
2.4. Processing Accident Information
Accidents that included unit types such equally cyclists, pedestrians, wheelchairs and animals were removed as these units are not afflicted by traffic in the same fashion as vehicles in the chief traffic stream.
The appointment-times of each accident were formatted into the ISO 8610 appointment-fourth dimension format with the Adelaide time-zone specified. Standardising date-times betwixt datasets will ensure the accurate temporal joining of accidents and traffic volume measurements.
Equally traffic volume data only exists for intersections within the ACC betwixt the years 2010 and 2014, the accidents were filtered to fit these parameters, leaving 2336 accidents (Table ane). Accident times were rounded down to the nearest hour to match the hourly timestamps of the traffic volume data. It was essential to round times downwardly to the previous hour to ensure the traffic book used was not affected by the blow itself. This do is used in previous studies looking at the human relationship between traffic volume and blow frequencies [34,forty,41].
Table ane
Accident information characteristics.
| Raw | Processed | |
|---|---|---|
| Spatial Extent | South Australia | ACC |
| Temporal Extent | 2010–2017 | 2010–2014 |
| northward | 146,718 | 2336 |
2.v. Intersection Traffic Volume Data
Traffic intersection volumes from 2010 to 2014 [42] are publicly available through data.sa.gov.au. The dataset consists of hourly traffic volume measurements for 122 intersections in the ACC; recorded using the Sydney Coordinated Adaptive Traffic Organisation (SCATS). Traffic volumes correspond the total number of vehicles to pass through an intersection in each hr. Directional traffic data was not available—subsequent methods detail the approach used to address this. Each hourly measurement includes the coordinates of its corresponding intersection, pregnant that every measurement at each intersection has a separate spatial data point. Over five meg hourly traffic volume measurements are available to be paired to individual accidents (Table 2). This is of import due to the rarity of blow events. A large traffic volume dataset increases the probability that any private blow will have associated traffic data and increases the number of accidents useable in the assay.
Table 2
Intersection traffic volumes data feature.
| Raw | Processed | |
|---|---|---|
| Spatial extent | ACC intersections | |
| Temporal extent | 2010–2014 | |
| Temporal resolution | sixty minutes | |
| Measurement resolution | one vehicle | |
| Number of intersections | 122 | 120 |
| north | 5,369,323 | 5,213,580 |
2.6. Processing Intersection Traffic Book Data
Upon investigation of the data, an intersection on Anzac Highway and i on Wakefield Street were plant to have median traffic volumes of zero vehicles per hour. Just over half of the traffic volume measurements at the Wakefield street site were zero and nearly all measurements at the Anzac Highway site were zero; this is unrealistic for ii major roads in the ACC. Volume measurements from these two intersections were removed from the dataset, leaving a total of 120 intersections (Table 2). There were big groups of consecutive zippo vehicles per 60 minutes readings—frequently during hours of the day where volumes above zero would exist expected—these are likewise errors. To address this, groups of traffic book measurements that remained the same for more than five consecutive hourly periods, including values to a higher place cipher, were removed. Overall, removing mistake measurements reduced the number of traffic volume measurements by approximately 150,000 (Tabular array 2).
Hour of solar day and appointment columns were combined into ane date-fourth dimension column and formatted in the ISO 8601 date-time format [43] with the Adelaide time-zone specified.
Traffic book measurements were also corrected using the provided error ratio, which indicates the proportion of vehicle counts in each hourly menses that were made in fault. The inverse of this ratio is the "valid ratio"; the proportion of vehicles that were counted correctly in whatsoever given hour. Each hourly traffic volume measurement was multiplied past its valid ratio to requite a corrected measurement, accounting for mistake in the SCATS sensors. The SCATS system is developed by the New South Wales Government in Commonwealth of australia and uses information collected from detectors at intersections to manage traffic signals.
2.7. Joining Accident and Traffic Volume Datasets
Earlier assay of the effects of traffic volumes on the frequency of accidents could be conducted, it was necessary to know the volume of traffic passing through an intersection immediately earlier each accident. This required joining the accident and intersection traffic volume datasets.
Using the coordinates of each accident and each traffic volume measurement, the two datasets were spatially joined with a altitude parameter of 20 grand; joining each blow to the traffic volume data for whatever intersection within 20 m. Because every hourly traffic volume measurement has its own spatial data indicate, each accident tape was duplicated across every traffic volume measurement at its intersection. This big dataset was filtered to only include rows where the engagement-time of the blow matched with the date-time of the traffic volume measurement. This resulted in a full of 1629 accidents (Table 3) with associated traffic volumes. This new table will be referred to equally the accident volumes dataset.
Tabular array 3
Characteristics of the created accident volumes dataset.
| Accident volumes | |
|---|---|
| Spatial extent | ACC intersections |
| Temporal extent | 2010–2014 |
| n | 1629 |
The accidents in this dataset were and then used to analyse the furnishings of traffic volumes on the occurrence of accidents.
2.8. Rainfall Information
High temporal resolution rainfall data was purchased from the Bureau of Meteorology (BOM) [44]. Data for the "Adelaide (Kent Town)" rainfall station was available from 1995 to 2015, with a total of 353,439 measurements. Rainfall rates were taken in increments of 0.two mm with a temporal resolution of thirty minutes.
2.ix. Accounting for Variability in Intersection Capacity
Every bit each intersection has a dissimilar capacity, traffic volumes are not comparable betwixt them. For instance, 1000 vehicles per hour may be shut to the capacity of one intersection but hands inside the chapters of another.
To business relationship for this, traffic volumes must be normalised. Traditionally, 5/c ratios are used, with methods derived from the Highway Capacity Manual [36]. If this method were to be used, the capacity at signalized intersections would exist calculated individually for dissimilar lane groups using their capacity, adjusted saturation flow charge per unit, effective green traffic bespeak ratio and cycle length [36]. Still, the broad spatial scale of the study area makes the use of this method difficult. Signal timing data for each intersection was not easily accessible and intersection geometry information would have been difficult to ascertain and use over 120 intersections. Additionally, traffic book data was simply available equally a total count for each intersection. The lack of directional vehicle counts makes the calculation of v/c ratios for dissimilar lane groups impossible. As a result, a novel approach to standardising traffic weather was taken.
This was achieved by assigning each measurement into one of xv bins in a quantile classification based on other measurements at the aforementioned intersection (Appendix A explains the selection of 15 bins). A traffic volume measurement of 300 vehicles per hour may exist assigned to bin fifteen at a depression-volume intersection, while a measurement of 5000 vehicles per hour may be assigned to bin xv at a loftier-book intersection. Looking at the two traffic volumes lone, they seem incomparable; however, they both fall among the highest volume measurements for their respective intersections. These bins effectively act equally an alphabetize for congestion by representing traffic volumes relative to the overall range of volumes at an intersection.
2.ten. Analysing the Human relationship Between Traffic Book and Accident Frequency
Intersections were grouped into iii different sizes based on their median traffic volumes. Accidents in the accident volumes dataset were and so grouped by the size of the intersection they occurred at and the congestion level at the time of the accident. This results in 45 groups (three intersection size ranks × fifteen congestion levels). The number that occurred in each of these 45 groups was counted.
Even so, plotting accident frequencies against congestion index on a linear scale results in a transformation of the response of blow frequency. This is considering the 15 congestion levels are not distributed evenly throughout the traffic volume distributions at intersections. To combat this, the median traffic volumes of each of the congestion levels was calculated, as explained in Appendix B.
Accident frequencies were then able to be plotted against this median value, allowing the linear hypothesis to be tested.
To decide whether the response of accident frequency to traffic volume was linear, or whether a significant not-linear effect was present, various models were fitted to blow frequency for each intersection size. As the data is non-negative count data, regular linear models are not advisable.
Initially, poisson generalized linear models (GLM) were fit with a single linear explanatory term. These models were then tested for overdispersion to make up one's mind whether the poisson was appropriate. If the poisson model is overdispersed, the negative binomial model is more advisable. The following formulae were used for either the poisson or negative binomial models.
- Linear: blow frequency ~ traffic volume
- Quadratic: accident frequency ~ traffic book + (traffic book)ii
- NaturalSpline: accident frequency ~ natural spline (traffic volume, 4 d.f.)
The most preferable of these three models for each intersection rank were determined using the AICc (Akaike Information Criterion) model selection criterion [45].
two.eleven. Accident Severity Analysis
For assay of how congestion affects accident severity, the accident volumes dataset was filtered into three subsets, containing property impairment only (PDO), minor injury (MI) and serious injury (SI) accidents (in that location were no fatal accidents at intersections in the ACC during the study period). As there were only 20 SI accidents with traffic volume data, at that place was too much racket for a articulate response to be observed and SI accidents were not considered further.
The frequency of PDO and MI traffic accidents in each congestion level were then plotted, allowing whatever difference in response of accident frequency to be seen. Normalised frequencies are the proportion of total PDO or MI accident counts in each congestion level. The accident frequency ratio is the ratio of PDO to MI frequencies in each congestion level.
two.12. Rainfall Take a chance Analysis
To decide the effect of rain on blow occurrence, the accidents in the accident volumes tabular array were separated into accidents occurring while it was raining and while information technology was not raining. For each of these filtered datasets, the accident frequency in each congestion level was counted.
Using these not-raining and raining blow frequencies, the risk of not-raining and raining accidents in each congestion level were calculated. Accident adventure is the probability of an accident occurring within a flow. Using raining accident risk as an example, risk was calculated using Equation (1):
Raining accident risk = Total # of accidents while raining/Total # of periods in which information technology was raining,
(1)
where a period refers to the hourly traffic volume periods.
To understand how rainfall risk changes with increasing congestion, this calculation was applied to each of the 15 congestion levels (Equation (2)).
Raining accident run a risk (Cx) = Full # of Cx accidents while raining/Total # of Cx periods in which it was raining,
(2)
where Cx is the congestion level (1–15).
The process was repeated for not-raining accidents.
To determine the total number of hourly periods in which information technology was raining (or not raining), the BOM rainfall information was joined to the intersection volumes table. The number of traffic book measurement periods in which it was raining and not raining were counted for each of the xv levels of congestion, assuasive risks to be calculated for each level.
Relative gamble (RR) is the ratio of the adventure of an event occurring nether exposed weather condition to the chance of an event occurring under control conditions [46]. In the context of accident risks, RR was calculated for each congestion level (Cx) using the following equation and the chance values calculated using Equations (1) and (2).
Relative risk (Cx) = Raining Blow Adventure (Cx) / Non-Raining Accident Risk (Cx),
(three)
RR was and so plotted against congestion index to allow any changes with increasing congestion to be observed. A change in RR would indicate a change in how rainfall affects the risk of an accident. A RR of greater than i means that the hazard of an blow occurring is higher when it is raining, while a RR of less than one means that the chance of an blow occurring is greater while information technology is not raining.
iii. Results
iii.ane. Relationship Between Traffic Volume and Accident Frequency
Based on Table 4, poisson GLMs were advisable for the counts of accidents occurring in low-volume intersections. As the poisson is overdispersed for centre- and high-volume intersections, negative binomial models were used for these instead.
Table 4
Dispersion statistics for poisson models fitted to data for low, centre- and loftier-book intersections. Dispersion ratios higher up one indicate possible overdispersion of information. A p-value of less than 0.05 indicates that the data is overdispersed.
| Intersection Rank | Dispersion Ratio | Pearson's Chi2 | p-Value | Overdispersed |
|---|---|---|---|---|
| Low-volume | ane.37 | 17.82 | 0.164 | No |
| Middle-volume | three.77 | 48.99 | <0.001 | Yes |
| High-volume | iii.25 | 42.25 | <0.001 | Aye |
Considering the rule of pollex that a delta AIC of 2 or more than gives substantial back up for the highest-ranked model [47], the quadratic models are favourable for blow counts at low- and middle-volume intersections, with evidence ratios of 17.vii and 11.5, respectively (Table v). The delta AICc between the quadratic and natural spline negative binomial models was only 1.2 for high-volume intersections (Table v), and so the AICc values do not back up the choice of 1 model over the other. However, a delta AICc of 7.viii for the model using only a linear explanatory term indicates a significantly meliorate fit of the two non-linear models.
Table 5
Model selection statistics for models fit to accident information for depression, middle- and loftier-volume intersections.
| Model | d.f. | Log Lik. | AICc | Delta AICc | Weight | Evidence Ratio |
|---|---|---|---|---|---|---|
| Low-book intersections | ||||||
| Quadratic | iii | −30.57 | 69.iii | - | 0.938 | 17.seven |
| Natural spline | five | −29.21 | 75.1 | 5.eight | 0.053 | - |
| Linear | two | −36.82 | 78.half-dozen | 9.3 | 0.009 | - |
| Center-book intersections | ||||||
| Quadratic | four | −48.05 | 108.1 | - | 0.912 | xi.5 |
| Natural spline | half dozen | −45.25 | 113.0 | four.ix | 0.079 | - |
| Linear | 3 | −54.57 | 117.3 | ix.two | 0.009 | - |
| High-volume intersections | ||||||
| Quadratic | four | −53.55 | 119.1 | - | 0.634 | 1.80 |
| Natural spline | 6 | −48.89 | 120.iii | one.2 | 0.352 | - |
| Linear | 3 | −59.34 | 126.9 | 7.8 | 0.013 | - |
Figure 3 emphasizes the linearity of the relationship up until the higher levels of congestion, with linear regressions plumbing fixtures within the 95% conviction bands of the loess curves up until this betoken. For middle- and high-volume intersections, the human relationship is linear up until median traffic volumes relating to congestion level 13. For low-volume intersections, the relationship is linear upwardly until median traffic volumes relating to congestion level 12. This suggests that the statistically meaning not-linear result (Tabular array 5) is largely a result of an exacerbated increase in accident frequency in highly congested weather condition. Appendix C shows the loess regressions that the conviction bands in Figure 3 are based on.
Relationship between traffic volume (median of each congestion level) and accident frequency. The dashed lines are linear regressions. For low-volume intersections, the linear regression is only fit for the median traffic volumes of the first 12 congestion levels. For middle- and high-volume intersections, the linear regression is fit for median volumes of the get-go xiii congestion levels. 95% confidence intervals chronicle to loess regressions fit to the same information.
3.2. Accident Severity
Slight differences in the response of normalised accident counts to congestion between the ii severities can be seen in the loess curves (Figure fourA); still, there is likely little significance to this observation, with no change in the ratio of MI to PDO accidents existence apparent (Figure fourB).
Effect of congestion on property damage merely (PDO) and minor injury (MI) accidents: (A) Response of PDO and MI accident frequencies to congestion alphabetize. The y-axis shows normalised accident counts. The curves are loess regressions; (B) Change in the ratio of MI and PDO accidents with increasing congestion index level. Conviction band is 95%.
3.3. Rainfall Take a chance
Accident run a risk is the probability that an accident will occur in whatever item hour. Looking at congestion level 15 in Effigy fiveA, for example, the risk of approximately 0.0008 for not-raining accidents means that, while it is non raining, there is a 0.08% chance of an accident occurring at an intersection in the ACC in this congestion level. For both not-raining and raining accidents, the risk of an blow occurring increases with increasing congestion (Figure 5A).
(A) Non-raining and raining accident risks; (B): Relative risk between not-raining and raining accident risks. Confidence band is 95%.
While raining risks remain higher than not-raining risks (Figure 5A), the RR becomes smaller (Figure 5B) as congestion increases. In congestion level one, a RR of approximately 5 means that the chance of an accident is five times greater when it is raining than when information technology is not raining. By congestion level fifteen, the RR approaches one, which would bespeak that the run a risk of an accident occurring while it is raining is equal to the risk of an blow occurring while it is not raining. The aforementioned conclusion can be made when looking at normalised raining and non-raining accident counts (Appendix D, Figure A6A) and the change in the ratio of raining to not-raining accident counts (Appendix D, Figure A6B).
four. Discussion
four.1. Relationship Betwixt Traffic Volume and Blow Frequency
Results present a non-linear quadratic response between traffic volume and accident frequency in depression, centre- and high-book intersections, further increasing support for the non-linear relationship between the ii variables. While the correlation is positive, as presented in about publications finding linear responses [24,25,26,27,28], the quadratic explanatory term was meaning, indicating a more complex relationship. The non-linear quadratic response supports the findings of Dickerson, et al. [48] for the response of blow frequencies to traffic menstruation, equally well as the results of Gwynn [29], Zhou and Sisiopiku [49] and Martin [31] when considering accident rates. The quadratic response emphasizes the do good reducing congestion could accept on blow occurrences. This is because reducing traffic volumes from the highest level would consequence in steep decline in accident occurrences compared to the aforementioned reduction at lower volumes where accident frequency increases more than linearly. Comparatively, a concave relationship would point a reduced effectiveness of decreasing traffic volumes from the highest levels.
The power of this analysis to detect the presence of a quadratic response may come down to several reasons.
Firstly, the quality of the traffic data used. Studies reviewed by Retallack and Ostendorf [22] using ii–three million hourly traffic measurements reported concave non-linear responses. The 5.2 million data points used in this study represent a significant increase, supporting the observation that large amounts of high-quality data allow more detailed relationships to be uncovered [22]. High temporal resolution traffic data allows the precise identification of traffic conditions at the time of each accident. As accidents are rare events, obtaining large sample sizes of accidents with respective traffic volume data is hard [50]. Large traffic volume datasets increment the likelihood that an accident can exist paired with volume data, increasing the last sample size of accidents useable for analysis. Hossain et al. [fifty] annotation that only 30 studies in their comprehensive review had sample sizes larger than 500. Comparing this to n = 1629 in this report highlights another point where additional detail was able to be captured.
Additionally, the increased homogeneity of the highly localized written report surface area reduces the potential of covariates to add noise to the relationship. Analysing data taken from a heterogenous written report surface area could result in the loss of item on a smaller scale, instead providing an averaged relationship for the surface area as a whole. This effect was observed past Dickerson et al. [48], where the analysis of information from four different road classes produced a linear relationship between traffic volume and accidents. When accidents were separated by road course, a non-linear effect was identified in the upper traffic volumes [48]. Similarly, Sullivan [51] was able to find the specific effects of queuing on accident occurrence when using disaggregated data, while aggregated data only uncovered a potential effect of congestion on blow occurrence.
Other variables such every bit vehicle speeds and speed variation may also exist considered to influence accident occurrence. Yet, this may be less relevant when looking at congested conditions in intersections. In not-congested conditions on direct route segments, the speeds of vehicles would accept a greater influence on the chance of an accident occurring compared to congested weather in intersections where traffic is non free-flowing and it is the number of vehicles present which has the greater effect. Although we have estimated congestion based on traffic volumes prior to accidents, secondary accidents may occur. This scenario may exist addressed with more than involved statistical methods using more covariates.
We find the highest rate of increment in accident frequencies at the highest congestion levels. Hence, targeting intersections that regularly reach the highest levels of congestion may achieve the all-time results in terms of traffic blow management. This coincides with general benefit of reducing congestion to increment productivity and reduce delays, pollution, and stress [52].
iv.2. Blow Severity
The lack of a clear deviation of how congestion effects the occurrence of PDO or MI accidents differs from the results of Mussone, et al. [53], who found traffic volumes to be able to predict the severity of accidents. Similarly, Abdel-Aty and Keller [54] plant the ADT of roads inbound intersections to be significant in predicting no-injury, possible injury and incapacitating injury accidents. The lack of difference in the relationship between MI and PDO accidents may come from the low maximum vehicle speeds in the written report area; with virtually all intersection sites having speed limits of 50 km/h. The protection provided by the vehicle may buffer any potential influence of congestion at these relatively depression speeds. If accidents including vulnerable unit types such as pedestrians and cyclists were not excluded, a difference in accident severity between congestion levels could have been found, equally these unit of measurement types are more than susceptible to injury.
Wang et al. [21] found congestion to have a significant influence on the occurrence of fatal and SI accidents but not MI accidents. This could indicate another reason for the lack of effect of congestion on blow severity; SI and fatal accidents were excluded due to limited sample sizes (n = 20 and n = 0, respectively).
four.3. Rainfall Risk
Many by studies looking at the effects of rainfall on accident risk apply the matched-pair method for calculating relative hazard [9,55,56,57,58,59,60]. In this approach, time periods are paired in a mode that one of the periods is wet, and the other is dry. Periods are from the same fourth dimension of solar day ane week apart. The "relative chance" is then considered to be the total number of accidents in the wet periods, divided by the number of accidents in dry periods. This differs from the method used in this study, where the RR is the ratio of ii proportions, rather than the ratio of raining and not-raining blow counts. The matched-pair method would not be possible for calculating RRs across the 15 congestion index levels established in our analysis. As levels of congestion are continually changing, it would be extremely difficult to find pairs of wet and dry fourth dimension periods from one week to the next that besides take the aforementioned congestion level.
Results propose that in college congestion levels the congestion itself is a more significant contributing gene to accident occurrence; with rainfall having less of an effect. This could provide insight for improved interpretation of the RR results in past studies. Hambly et al. [58], for example, investigated the wider Vancouver, Canada, region, finding a RR of ane.22. It is possible that this value is not indicative of risks in sections of the report area where high levels of congestion are common. Our results propose that this value may exist an overestimate in congested locations.
v. Conclusions
This study has demonstrated the power of high temporal frequency traffic book information to be used in parsimonious models for predicting accident frequencies at intersections. A total of 1629 motor vehicle accidents were linked traffic volume data from a pool of over five million hourly traffic volume data points.
Results show that accident frequency increases non-linearly in the college levels of congestion. Therefore, suggesting that managing traffic to avoid such loftier levels of congestion would have the greatest impact on reducing accident occurrence. Importantly, there is no appreciable increase in blow frequency equally congestion decreases, meaning that reducing congestion would not negatively affect public wellness.
Change in the severity of accidents between congestion levels was likewise considered. Withal, no human relationship was found, perchance due to the lack of SI and fatal accidents in the data.
Rainfall risks were compared individually for each of the fifteen levels of congestion, showing an increased influence of rainfall on accident occurrence when levels of congestion are low and indicating an increased importance of rainfall adventure management in these weather.
This assay has demonstrated the benefit of using long-term, wide-calibration, temporally detailed data for accident take chances analysis.
Acknowledgments
Acknowledgements to the Department of Planning, Transport and Infrastructure for providing blow data for employ in this research; and to Dr Steven Delean for his assistance.
Appendix A. Determining the Platonic Number of Traffic Volume Bins
The Freedman–Diaconis rule [61] was used to assist in determining the ideal number of bins to divide traffic volumes into. This dominion specifies that the ideal bin width is given by Equation (A1):
(A1)
where x is the traffic volumes of each of the 1629 accidents and n is the number of traffic accidents.
Because blow frequencies were plotted separately for each of the three intersection traffic volume ranks, the ideal bin width was calculated for each of these groups separately. Using accidents in depression-volume intersections for case:
So, traffic volumes for accidents occurring in low-volume intersections should ideally be divided into bins with widths of approximately 282 vehicles per hour. The ideal number of bins for this distribution was then calculated by dividing the total range of traffic volumes of crashes at low-volume intersections by the platonic bin width (Equation (A2)):
(A2)
This process was repeated for accidents occurring at eye- and high-volume intersections. The results are shown in Table A1.
Table A1
The ideal number of bins for accidents in low, middle- and high-volume intersections calculated using the Freedman–Diaconis rule.
| Intersection Rank | Platonic Bin Width | Ideal number of Bins |
|---|---|---|
| Low-volume | 281.995 | 9.71 |
| Centre-volume | 299.131 | xvi.69 |
| High-volume | 426.830 | 17.62 |
| All | 320.916 | 23.46 |
Based on these results, traffic volumes were divided into 10, 15, 20 and 25 bins.
However, for ease of understanding, the number of bins should remain constant throughout the report. To decide which number to choose, accident frequencies were plotted against the median traffic volumes of each of these bins, equally explained in Appendix B.
Effigy A1 shows the fit of loess regressions to each of the intersection book ranks when traffic volumes are divided into x, 15, 20 and 25 bins.
While the curve for accidents in low-volume intersections fits well when x bins are used, the bend for heart-book intersections appears to be overly affected past noise. When volumes are divided into 25 bins, four of the lowest traffic volume bins comprise zero accidents; using fewer bins reduces the chances of bins containing no accidents. Comparing the employ of fifteen and 20 bins there is little difference to the fit of the loess visually, so xv bins will be used as this is closer to the ideal number of bins for accidents in low-book intersections (Table A1).
Figure A1
Comparison of traffic volumes grouped into 10, xv, 20 and 25 bins. Accident frequencies are plotted against the median traffic volume of each bin, and separate loess curves are fit for accidents at low, middle- and loftier-volume intersections.
Appendix B. Computing the Median Traffic Volumes of Congestion Levels
Figure A2 shows how the 15 congestion levels relate to the traffic volume distributions of the highest, middle and lowest median traffic book intersections. At the two highest volume intersections, the bins are distributed relatively evenly across the range of traffic volumes; however, in lower volume intersections the distribution becomes positively skewed. This is considering high traffic volumes are rare, and as each bin contains the same number of traffic volume measurements, the bin width increases. This uneven spread of bins causes an issue when investigating the response between traffic book and accident frequency. If blow frequencies were plotted against the 15 congestion levels on a linear scale, despite the not-linear spread seen in Effigy A2, the upper terminate of the scale is compressed. This exaggerates the increase in accident frequency (Figure A3).
Figure A2
Traffic volume density distributions of the top, middle and bottom two intersection sites, ordered by median traffic volume. The lines represent where the 15 bins sit in the distribution.
Figure A3
Blow frequencies plotted against the 15 traffic book bins.
When the same blow frequency values are plotted against the median traffic volumes of the 15 bins Figure A4), the distribution of data points better represents the distribution of bins shown in Figure A2, and the response of blow frequencies is no longer afflicted.
Figure A4
Accident frequencies plotted confronting the median volumes of the 15 bins.
Notwithstanding, information technology is important to note that taking the median traffic volumes of each of the 15 congestion index levels from all intersections will crusade problems. This is due to the differences in the distribution of traffic volumes at each intersection. Consider if the median traffic volume of congestion level 15 is taken from all intersections. This volume will likely sit well below the maximum volume of a high-book intersection and above the maximum seen at a low-volume intersection. To accost this, intersections are divided into three ranks based on their median traffic volumes. The median traffic volumes of each congestion level can and so be calculated separately from intersections of each rank.
Appendix C. Effect of Traffic Volume on Accident Frequency
Figure A5
Loess curves showing the response of accident frequency to changing congestion in different sized intersections. Error bands are at a 95% confidence level.
Appendix D. Normalised Not-Raining and Raining Accident Frequencies
Figure A6
(A) Change in normalised not-raining and raining accident counts with increasing congestion; (B) Change in the ratio of raining accidents to not-raining accidents with increasing congestion. The fitted lines are loess regressions. Error bands are at a 95% conviction level.
Author Contributions
Conceptualization, A.E.R. and B.O.; methodology, A.E.R. and B.O.; software, A.E.R. and B.O.; visualization, A.E.R.; formal analysis, A.Eastward.R and B.O.; investigation, A.E.R.; data curation, A.E.R.; writing—Original typhoon training, A.E.R.; writing—Review and editing, A.E.R and B.O.; supervision, B.O. All authors take read and agreed to the published version of the manuscript.
Funding
This enquiry received no external funding.
Conflicts of Interest
The authors declare no conflict of interest.
References
ane. Litchfield F. The Price of Road Crashes in Australia 2016: An Overview of Safety Strategies. The Australian National University; Canberra, Australia: 2016. [Google Scholar]
2. Hensher D.A., Rose J.K., Ortúzar J.d.D., Rizzi L.I. Estimating the willingness to pay and value of risk reduction for machine occupants in the route surroundings. Transp. Res. Part A Policy Pract. 2009;43:692–707. doi: ten.1016/j.tra.2009.06.001. [CrossRef] [Google Scholar]
3. BITRE . Road Crash Costs in Commonwealth of australia 2006. Agency of Infrastructure, Transport and Regional Economic science; Canberra, Australia: 2009. pp. 1–110. [Google Scholar]
4. Evans L. Traffic Prophylactic and the Driver. Van Nostrand Reinhold; New York, NY, USA: 1991. [Google Scholar]
5. BITRE . Impact of Road Trauma and Measures to Improve Outcomes. Department of Infrastructure and Regional Development; Canberra, Commonwealth of australia: 2014. [Google Scholar]
6. BITRE . Road Trauma Australia 2018 Statistical Summary. Bureau of Infrastructure, Transport and Regional Economic science; Canberra, Australia: 2019. [Google Scholar]
vii. Andreescu M.-P., Frost D.B. Weather and traffic accidents in montreal, canada. Clim. Res. 1998;ix:225–230. doi: ten.3354/cr009225. [CrossRef] [Google Scholar]
viii. Eisenberg D. The mixed furnishings of precipitation on traffic crashes. Accid. Anal. Prev. 2004;36:637–647. doi: 10.1016/S0001-4575(03)00085-X. [PubMed] [CrossRef] [Google Scholar]
ix. Keay K., Simmonds I. Road accidents and rainfall in a large australian urban center. Accid. Anal. Prev. 2006;38:445–454. doi: 10.1016/j.aap.2005.06.025. [PubMed] [CrossRef] [Google Scholar]
ten. Sherretz L.A., Farhar B.C. An analysis of the human relationship betwixt rainfall and the occurrence of traffic accidents. J. Appl. Meteorol. 1978;17:711–715. doi: 10.1175/1520-0450(1978)017<0711:AAOTRB>2.0.CO;2. [CrossRef] [Google Scholar]
11. Veh A. Improvements to Reduce Traffic Accidents. Meeting of the Highway Division; New York, NY, USA: 1937. pp. 1775–1785. [Google Scholar]
12. Ivan J.N., Wang C.Y., Bernardo N.R. Explaining two-lane highway crash rates using land apply and hourly exposure. Accid. Anal. Prev. 2000;32:787–795. doi: x.1016/S0001-4575(99)00132-iii. [PubMed] [CrossRef] [Google Scholar]
13. Shefer D., Rietveld P. Congestion and safety on highways: Towards an analytical model. Urban Stud. 1997;34:679–692. doi: x.1080/0042098975970. [CrossRef] [Google Scholar]
xiv. Lie A., Tingvall C., Krafft M., Kullgren A. The effectiveness of electronic stability control (esc) in reducing real life crashes and injuries. Traffic Inj. Prev. 2006;7:38–43. doi: ten.1080/15389580500346838. [PubMed] [CrossRef] [Google Scholar]
15. Iversen H., Rundmo T. Personality, risky driving and accident interest amidst norwegian drivers. Personal. Ind. Differ. 2002;33:1251–1263. doi: 10.1016/S0191-8869(02)00010-vii. [CrossRef] [Google Scholar]
16. Horwood L.J., Fergusson D.1000. Potable driving and traffic accidents in young people. Accid. Anal. Prev. 2000;32:805–814. doi: x.1016/S0001-4575(00)00005-1. [PubMed] [CrossRef] [Google Scholar]
17. Washington S., Metarko J., Fomunung I., Ross R., Julian F., Moran Eastward. An inter-regional comparing: Fatal crashes in the southeastern and non-southeastern united states of america: Preliminary findings. Accid. Anal. Prev. 1999;31:135–146. doi: ten.1016/S0001-4575(98)00055-iv. [PubMed] [CrossRef] [Google Scholar]
18. O'donnell C., Connor D. Predicting the severity of motor vehicle accident injuries using models of ordered multiple choice. Accid. Anal. Prev. 1996;28:739–753. doi: ten.1016/S0001-4575(96)00050-4. [PubMed] [CrossRef] [Google Scholar]
19. Shankar V., Mannering F., Barfield W. Outcome of roadway geometrics and environmental-factors on rural freeway accident frequencies. Accid. Anal. Prev. 1995;27:371–389. doi: x.1016/0001-4575(94)00078-Z. [PubMed] [CrossRef] [Google Scholar]
20. Milton J., Mannering F. The human relationship among highway geometrics, traffic-related elements and motor-vehicle accident frequencies. Transportation. 1998;25:395–413. doi: 10.1023/A:1005095725001. [CrossRef] [Google Scholar]
21. Wang C., Quddus M., Ison S. A spatio-temporal assay of the impact of congestion on traffic condom on major roads in the uk. Transp. A Transp. Sci. 2013;9:124–148. doi: ten.1080/18128602.2010.538871. [CrossRef] [Google Scholar]
22. Retallack A.Eastward., Ostendorf B. Current agreement of the effects of congestion on traffic accidents. Int. J. Environ. Res. Public Wellness. 2019;16:3400. doi: 10.3390/ijerph16183400. [PMC free article] [PubMed] [CrossRef] [Google Scholar]
23. Raff M.Due south. Interstate highway—Accident study. Highw. Res. Board Balderdash. 1953;74:18–45. [Google Scholar]
24. Woo J.C. Correlation of Accident Rates and Roadway Factors. Purdue Academy; Lafaytette, IN, USA: 1957. pp. 2326–6325. [Google Scholar]
25. Schoppert D.W. Predicting traffic accidents from roadway elements of rural two-lane highways with gravel shoulders. Highw. Res. Board Bull. 1957;158:4–26. [Google Scholar]
26. Head J.A. Predicting traffic accidents from roadway elements on urban extensions of land highways. Highw. Res. Board Bull. 1959;208:45–63. [Google Scholar]
27. Cadar R.D., Boitor M.R., Dumitrescu M. Effects of traffic volumes on accidents: The case of romania'due south national roads. Geogr. Tech. 2017;12:twenty–29. [Google Scholar]
28. Vitaliano D.F., Held J. Road accident external furnishings: An empirical assessment. Appl. Econ. 1991;23:373–378. doi: 10.1080/00036849100000146. [CrossRef] [Google Scholar]
29. Gwynn D.W. Relationship of accident rates and blow involvements with hourly volumes. Traffic Q. 1967;21:407–418. [Google Scholar]
30. Ceder A. Relationships betwixt road accidents and hourly traffic flow—ii: Probabilistic approach. Accid. Anal. Prev. 1982;14:35–44. doi: ten.1016/0001-4575(82)90005-7. [CrossRef] [Google Scholar]
31. Martin J.-L. Relationship between crash rate and hourly traffic period on interurban motorways. Accid. Anal. Prev. 2002;34:619–629. doi: 10.1016/S0001-4575(01)00061-6. [PubMed] [CrossRef] [Google Scholar]
32. Frantzeskakis J.Chiliad., Iordanis D.I. Volume-to-capacity ratio and traffic accidents on interurban four-lane highways in greece. Transp. Res. Rec. 1987;1112:29–38. [Google Scholar]
33. Shefer D. Congestion, air-pollution, and road fatalities in urban areas. Accid. Anal. Prev. 1994;26:501–509. doi: ten.1016/0001-4575(94)90041-8. [PubMed] [CrossRef] [Google Scholar]
34. Theofilatos A. Incorporating existent-time traffic and weather data to explore road blow likelihood and severity in urban arterials. J. Saf. Res. 2017;61:9–21. doi: 10.1016/j.jsr.2017.02.003. [PubMed] [CrossRef] [Google Scholar]
35. Mannering F., Bhat C.R. Analytic methods in accident research: Methodological frontier and future directions. Anal. Methods Accid. Res. 2014;1:one–22. doi: 10.1016/j.amar.2013.09.001. [CrossRef] [Google Scholar]
36. Transportation Research Board . Highway Capacity Manual—A Guide for Multimodal Mobility Analysis. 6th ed. Transportation Research Board; Washington, DC, The states: 2016. [Google Scholar]
37. R Core Team . R: A Language and Environment for Statistical Calculating. R Foundation for Statistical Computing; Vienna, Austria: 2019. [Google Scholar]
38. RStudio Team . Rstudio: Integrated Development for r. RStudio, Inc.; Boston, MA, USA: 2018. [Google Scholar]
39. DPTI . Road Crash Data. DPTI; Adelaide, South Australia: 2019. [Google Scholar]
xl. Yuan J., Abdel-Aty M. Approach-level real-time crash risk analysis for signalized intersections. Accid. Anal. Prev. 2018;119:274–289. doi: ten.1016/j.aap.2018.07.031. [PubMed] [CrossRef] [Google Scholar]
41. Ahmed M.M., Abdel-Aty M.A. A data fusion framework for real-fourth dimension risk assessment on freeways. Transp. Res. Role C Emerg. Technol. 2013;26:203–213. doi: 10.1016/j.trc.2012.09.002. [CrossRef] [Google Scholar]
42. City of Adelaide . Traffic Intersection Volumes. City of Adelaide; Adelaide, Australia: 2017. [Google Scholar]
44. Bureau of Meteorology . Rainfall—Total Reported on One-half Hourly: Adelaide (Kent Town) Bureau of Meteorology; Adelaide, South Australia: 2019. [Google Scholar]
45. Akaike H. Information theory and an extension of the maximum likelihood principle. In: Parzen Due east., Tanabe One thousand., Kitagawa G., editors. Selected Papers of Hirotugu Akaike. Springer; New York, NY, USA: 1998. pp. 199–213. [Google Scholar]
46. Tenny S., Hoffman 1000.R. Statpearls [Internet] StatPearls Publishing; Treasure Island, FL, The states: 2019. Relative take a chance. [Google Scholar]
47. Burnham Chiliad.P., Anderson D.R. Multimodel inference: Understanding aic and bic in model option. Sociol. Methods Res. 2004;33:261–304. doi: 10.1177/0049124104268644. [CrossRef] [Google Scholar]
48. Dickerson A., Peirson J., Vickerman R. Route accidents and traffic flows: An econometric investigation. Economica. 2000;67:101–121. doi: 10.1111/1468-0335.00198. [CrossRef] [Google Scholar]
49. Zhou M., Sisiopiku V.P. Relationship between volume-to-capacity ratios and blow rates. Transp. Res. Rec. 1997;1581:47–52. doi: ten.3141/1581-06. [CrossRef] [Google Scholar]
50. Hossain Chiliad., Abdel-Aty 1000., Quddus Yard.A., Muromachi Y., Sadeek South.N. Existent-time crash prediction models: Country-of-the-art, design pathways and ubiquitous requirements. Accid. Anal. Prev. 2019;124:66–84. doi: 10.1016/j.aap.2018.12.022. [PubMed] [CrossRef] [Google Scholar]
51. Sullivan E.C. Estimating accident benefits of reduced expressway congestion. J. Transp. Eng. 1990;116:167–180. doi: 10.1061/(ASCE)0733-947X(1990)116:2(167). [CrossRef] [Google Scholar]
52. Rao A.G., Rao K.R. Measuring urban traffic congestion-a review. Int. J. Traffic Transp. Eng. 2012;2:286–305. [Google Scholar]
53. Mussone 50., Bassani K., Masci P. Analysis of factors affecting the severity of crashes in urban road intersections. Accid. Anal. Prev. 2017;103:112–122. doi: 10.1016/j.aap.2017.04.007. [PubMed] [CrossRef] [Google Scholar]
54. Abdel-Aty M., Keller J. Exploring the overall and specific crash severity levels at signalized intersections. Accid. Anal. Prev. 2005;37:417–425. doi: 10.1016/j.aap.2004.11.002. [PubMed] [CrossRef] [Google Scholar]
55. Andrey J., Mills B., Leahy 1000., Suggett J. Weather condition every bit a chronic risk for road transportation in canadian cities. Nat. Hazards. 2003;28:319–343. doi: 10.1023/A:1022934225431. [CrossRef] [Google Scholar]
56. Andrey J. Long-term trends in weather condition-related crash risks. J. Transp. Geogr. 2010;18:247–258. doi: x.1016/j.jtrangeo.2009.05.002. [CrossRef] [Google Scholar]
57. Dai C. Exploration of Weather Impacts on Freeway Traffic Operations and Safe Using High-Resolution Weather Data. Portland State University; Portland, OR, Us: 2011. [Google Scholar]
58. Hambly D., Andrey J., Mills B., Fletcher C. Projected implications of climate alter for route safety in greater vancouver, canada. Clim. Chang. 2013;116:613–629. doi: ten.1007/s10584-012-0499-0. [CrossRef] [Google Scholar]
59. Sun X., Hu H., Habib E., Magri D. Quantifying crash risk under inclement weather with radar rainfall data and matched-pair method. J. Transp. Saf. Secur. 2011;3:ane–14. doi: 10.1080/19439962.2010.524348. [CrossRef] [Google Scholar]
60. Jaroszweski D., McNamara T. The influence of rainfall on route accidents in urban areas: A weather condition radar arroyo. Travel Behav. Soc. 2014;1:15–21. doi: ten.1016/j.tbs.2013.10.005. [CrossRef] [Google Scholar]
61. Freedman D., Diaconis P. On the histogram equally a density estimator: L2 theory. Probab. Theory Relat. Fields. 1981;57:453–476. [Google Scholar]
Manufactures from International Journal of Environmental Research and Public Health are provided hither courtesy of Multidisciplinary Digital Publishing Found (MDPI)
How To Calculate Number Of Accidents From Accident Raw Data Of Accident Id,
Source: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7068508/
Posted by: guitierrezbessithomfor.blogspot.com
0 Response to "How To Calculate Number Of Accidents From Accident Raw Data Of Accident Id"
Post a Comment